Fought v. UNUM Life Insurance Co. of America

PER CURIAM. Shirley O. Fought challenges the decision by UNUM’s claims administrator to deny long-term disability benefits under her employer’s group disability plan. A severe staph infection that followed elective heart surgery hospitalized and disabled Ms. Fought. UNUM’s plan administrator denied coverage by concluding that Ms. Fought suffered from a pre-existing coronary artery condition that “caused,” “contributed to,” or “resulted” in Ms. Fought’s disability, citing language in the plan. After exhausting the company’s internal appeals process, Ms. Fought brought a civil suit under 29 U.S.C. § 1132(a)(1)(B), alleging that she was entitled to disability benefits under the plan. *1177 UNUM admitted to a conflict of interest, as both payor and administrator of the plan. The magistrate judge denied discovery regarding the extent of UNUM’s conflict of interest. The district court then granted summary judgment in favor of UNUM. Exercising jurisdiction under 28 U.S.C. § 1291, we hold that the district court did not apply the appropriate standard of review when it considered the plan administrator’s denial of benefits to Ms. Fought. Applying the correct standard of review, we reverse the grant of summary judgment in favor of UNUM and remand to the district court for further proceedings. I. BACKGROUND A. Undisputed Facts On May 18, 1998, Ms. Fought enrolled in her employer’s group long-term disability plan, which was issued by UNUM with an effective date of June 1, 1998. The policy, under a provision entitled “What disabilities are not covered under your plan?” states: “Your plan does not cover any disabilities caused by, contributed to by, or resulting from your ... pre-existing condition.” Aple’s Supp.App. at 341-42. The policy does not define the terms “caused by, contributed to by, or resulting from.” The policy provides the following details regarding a pre-existing condition: You have a pre-existing condition when you apply for coverage when you first become eligible if: - you received medical treatment, consultation, care or services including diagnostic measure or took prescribed drugs or medicines in the 3 months just prior to your effective date of coverage; or you had symptoms for which an ordinarily prudent person would have consulted a health care provider in the 3 months just prior to your effective date of coverage; and - the disability begins in the first 12 months after your effective date of coverage. Id. at 342. Prior to her enrollment in the plan, Ms. Fought had been diagnosed and treated for coronary artery disease. In August 1998, approximately three months after her enrollment in the plan, Ms. Fought underwent angioplasty. On March 8,1999, she was admitted for unstable angina syndrome, and on March 15, 1999, she underwent an elective coronary artery revascu-larization surgery. During surgery, the doctors discovered that Ms. Fought’s sternum was narrow and osteoporotic, requiring a special procedure to close the surgical wound. Her doctors noted that “her postoperative course was anticipated to be quite challenging!,] given the concerns about the wound.” Aplt’s App. at 79. She was discharged six days after surgery on March 22, 1999. “At the time of [her] discharge, there [was] no evidence of infection,” and “her wounds were healing well.” Id. at 83; 166. Three weeks later, the incision from Ms. Fought’s wound became “dehisced,” or split open. On April 8, 1999, she was readmitted for care of her dehisced sternal wound and a possible infection. Aple’s Supp.App. at 97. At this time, her wound cultures tested positive for a “few” Klebsiella pheumonia bacteria. She was placed on antibiotics and given intensive wound care to prevent infection. After a hospital stay of five days, and a “dramatic improvement in the appearance of the wound,” she was sent to a skilled nursing facility. Aplt’s App. at 85. At the time, the “wound appearance looked satisfactory.” Id. at 85. She was discharged from the facility on April 19, 1999. On May 7, 1999, Ms. Fought complained of right-side chest pain. She was readmit *1178 ted to the hospital on May 11,1999, with a white blood cell count of 12,000 and a low grade fever. Two exposed sternal wires were detected. Her sternal wound and blood cultures were positive for both Kleb-siella pneumonia and methicillin-resistant Staphylococcus aureus in the sternal wound and methicillin-resistant Staphylococcus aureus in the blood stream. Id. at 139. Ms. Fought was placed in the Intensive Care Unit. Over the next two months, she underwent various operative procedures, was intubated, and received hemodynamic monitoring, nutritional support, and sedation. One surgery involved extensive ster-nal wound reconstruction and required Ms. Fought to be placed on a ventilator. She was discharged on July 15, 1999, when she was transferred to another facility for intensive wound care. On September 13, 1999, UNUM denied coverage under the long-term disability plan, having determined that Ms. Fought’s pre-existing condition “caused, contributed to, or resulted in the condition(s) for which [she was] claiming disability.” Aplt’s App. at 33-34 (Letter to Ms. Fought from Anne Dionne, Disability Benefit Specialist, dated Sept. 13, 1999). Ms. Fought submitted a formal request to have her claim reopened, see id. at 143 (Letter from Ms. Fought to Ann Dionne, dated Oct. 19, 1999), as well as letters from three doctors certifying that the staph infection was neither a preexisting condition nor related to her preexisting coronary artery disease. Id. at 156 (Note from Dr. Robert T. Ferraro, dated Oct. 19, 1999) (“[T]he staph infection which is the basis for multiple wounds on chest is not related to coronary artery disease. This is a separate, unrelated diagnosis without preceding history.”) (emphasis added); id. at 154 (Note from Dr. Robert Dubroff, dated Oct. 22, 1999) (“[Ms.] Fought is totally disabled due to her heart condition. The staph infection was not a pre-existing condition.”) (emphasis added); id. at 155 (Note from Dr. Neil T. Chen, undated) (“[Ms.] Fought’s chest & abdominal wounds/infection is [sic] not a preexisting condition.”) (emphasis in original). UNUM’s medical department reviewed Dr. Ferraro’s letter, 1 but the company declined to reverse its previous decision, stating that although “the staph infection itself was not present during the pre-existing condition period (3/1/98— 5/31/98), it was the result of surgery that was performed for a cardiac condition that was present, diagnosed and treated during that time frame.” Id. at 147-48 (Letter to Ms. Fought from Anne Dionne, dated Oct. 25, 1999). Ms. Fought then retained legal counsel, who contacted UNUM’s Long-Term Disability Quality Review Section, and in *1179 formed the company that Ms. Fought was appealing the denial of coverage. See id. at 159. After a review, UNUM again denied coverage, stating that “the staph infection was the result of surgery performed for a cardiac condition that was caused by, contributed to by, or resulted from the cardiac condition that was present, diagnosed and treated during the preexisting period.” Id. at 150 (Letter from John J. Schifano, Senior Benefit Analyst, dated Dec. 2,1999). Finally, Ms. Fought contacted the New Mexico Public Relations Commission, which corresponded with UNUM concerning Ms. Fought’s situation. Responding to the state agency’s inquiry, UNUM explained that the staph infection was the result of coronary bypass surgery, which was performed to treat her pre-existing condition. “She would not have had to have the surgery, later developing an infection, if she did not have the cardiac conditions which were present and treated for during the pre-existing period.” Id. at 158 (Letter from Theresa-Ann Uminga, Senior Complaints Specialist, to James A. Chavez, Ms. Fought’s Attorney, dated Feb. 11, 2000). B. Procedural History In August 2001, Ms. Fought filed suit in federal district court alleging UNUM violated 29 U.S.C. § 1132 in denying her claim for benefits. UNUM admitted that it operated under a conflict of interest, id. at 12 (Memorandum Opinion and Order Granting Defendant’s Motion for Summary Judgment, filed May 31, 2002), because it both administers claims and is the pay or of those claims. Ms. Fought requested discovery to allow inquiry into the extent of UNUM’s conflict of interest. The magistrate judge denied this request. UNUM moved for summary judgment. The district court, acknowledging the conflict of interest, proceeded to interpret the contract language “caused by, contributed to by, or resulting from” as meaning “related to” and “foreseeable complication of’ a pre-existing condition. The district court ruled that the “staph infection [was] related to the coronary artery disease as a foreseeable complication of treatment.” Aplt’s App. at 14. On that basis, the district court granted UNUM’s motion for summary judgment, concluding that UNUM’s decisions to deny Ms. Fought long-term disability benefits were not arbitrary or capricious, and ordered Ms. Fought to pay UNUM’s costs. Ms. Fought now appeals. II. ANALYSIS UNUM’s long-term disability plan is governed by the Employee Retirement and Income Security Act (ERISA), 29 U.S.C. § 1001 et seq. “ERISA was enacted to promote the interests of employees and their beneficiaries in employee benefit plans, and to protect contractually defined benefits.” Firestone Tire & Rubber Co. v. Bruch, 489 U.S. 101, 113, 109 S.Ct. 948, 103 L.Ed.2d 80 (1989) (citations and internal quotation marks omitted); see also 29 U.S.C. § 1001(b) (“It is hereby declared to be the policy of this chapter to protect ... the interests of participants in employee benefit plans and their beneficiaries ... by establishing standards of conduct, responsibility, and obligation for fiduciaries of employee benefit plans.”). In seeking coverage under her long-term disability benefit plan, Ms. Fought advances three arguments. First, she argues that the district court erred by using the wrong standard of review when it reviewed UNUM’s decision. Second, she argues that UNUM’s denial of benefits was, in any event, an unreasonable interpretation under the plan. Finally, she argues that the district court failed to consider UNUM’s obligation to draft plan provi *1180 sions in a manner calculated to be understood by the average plan participant. We begin with the appropriate standard of review, discussing (1) the “sliding scale” standard of review, (2) the two-tier approach to determining how far we “slide” upon that “scale,” and (3) the application of the reduced deference standard in this case. We then combine our analysis of Ms. Fought’s second and third arguments. Given our standard of review, we first analyze whether the plan administrator’s construction of the plan language is a reasonable one. In so doing, we consider (a) the plan’s pre-existing exclusion clause, (b) the role of causation in interpreting the pre-existing exclusion clause, (c) the Department of Labor’s regulations and a published example regarding pre-existing conditions, (d) relevant circuit and district court case law involving similar questions, and (e) whether clearer exclusionary language may have been available to UNUM. Finally, we examine whether, given the record evidence and our reduced deference standard of review, UNUM’s application of the pre-existing condition exclusion was supported by a preponderance of the evidence. A. The Standard of Review “Summary judgment orders are reviewed de novo, using the same standards as applied by the district court.” Pitman v. Blue Cross & Blue Shield of Okla., 217 F.3d 1291, 1295 (10th Cir.2000). Accordingly, like the district court, we must review UNUM’s decision to deny benefits to Ms. Fought, and we must determine the appropriate standard to be applied. The Supreme Court provided important guidance regarding the standard of review in ERISA benefits cases in Firestone, 489 U.S. at 113-15, 109 S.Ct. 948. The Court noted that deference to expert administrators is grounded in the most fundamental premises of trust law. If a disinterested party exercising discretionary powers has looked at evidence and rendered a decision, it is not only reasonable but a wise conservation of judicial resources not to have judges replicate the administrator’s work. See id. Recognizing that parties to a contract can agree to vest discretionary authority in an administrator, the Supreme Court held that “a denial of benefits challenged under § 1132(a)(1)(B) [ERISA] is to be reviewed under a de novo standard unless the benefit plan gives the administrator or fiduciary discretionary authority to determine eligibility for benefits or to construe the terms of the plan.” Id. at 115, 109 S.Ct. 948. There is no dispute that here the plan expressly gives UNUM, as plan administrator, the discretion to determine whether to deny a claimant insurance benefits under the plan. Aplt’s. App. at 31. Therefore, because the plan grants UNUM discretion, “[a] court reviewing a challenge to a denial of employee benefits ... applies an ‘arbitrary and capricious’ standard to a plan administrator’s actions.” Charter Canyon Treatment Ctr. v. Pool Co., 153 F.3d 1132, 1135 (10th Cir.1998). In reviewing a plan administrator’s decision under the arbitrary and capricious standard, “the federal courts are limited to the ‘administrative record’ — the materials compiled by the administrator in the course of making his decision.” Hall v. Unum Life Ins. Co. of Am., 300 F.3d 1197,1201 (10th Cir.2002). The possibility of an administrator operating under a conflict of interest, however, changes the analysis. Caldwell v. Life Ins. Co. of N. Am., 287 F.3d 1276, 1282 (10th Cir.2002) (“Indicia of arbitrary and capricious decisions include ... conflict of interest by the fiduciary.”). Thus, “if a benefit plan gives discretion to an administrator or fiduciary who is operating *1181 under a conflict of interest, that conflict must be weighed as a ‘factofr] in determining whether there is an abuse of discretion.’ ” Firestone, 489 U.S. at 115, 109 S.Ct. 948 (quoting Restatement (Seoond) of Trusts § 187, cmt. d (1959)). 2 “The rationale for this approach is clear. A conflicted fiduciary may favor, consciously or unconsciously, its interests over the interests of the plan beneficiaries.” Brown v. Blue Cross & Blue Shield, Inc., 898 F.2d 1556, 1565 (11th Cir.1990); see also Pit-man, 217 F.3d at 1296 (“ ‘[WJhen an insurance company serves as ERISA fiduciary ..., it is exercising discretion over a situation for which it incurs direct, immediate expense as a result of benefit determinations favorable to plan participants.’ ”) (quoting Brown, 898 F.2d at 1561). Following Firestone, the various circuit courts attempted to put the Court’s instructions into practice. “Since Firestone, all of the circuit courts agree that a conflict of interest triggers a less deferential standard of review. The courts, however, differ over how this lesser degree of deference alters their review process.” Chambers v. Family Health Plan Corp., 100 F.3d 818, 825 (10th Cir.1996). 1. Sliding Scale In Chambers, we identified two basic approaches that had emerged in interpreting Firestone: the “sliding scale” approach and the “presumptively void” approach. We explicitly adopted the former. Id. at 826-27. “Under [the sliding scale] approach, the reviewing court will always apply an arbitrary and capricious standard, but the court must decrease the level of deference given to the conflicted administrator’s decision in proportion to the seriousness of the conflict.” Id. at 825; see also Ladd v. ITT Corp., 148 F.3d 753, 754 (7th Cir.1998) (noting that when “the administrator has a conflict of interest, then, though the standard of review is nominally the same, the judicial inquiry is more searching”); Spangler v. Unum Life Ins. Co. of Am., 38 F.Supp.2d 952, 955-56 (N.D.Okla.1999) (noting that where conflict of interest is “apparent,” further discovery is not required; “the court’s review is a little more searching ... and the court is not as quick to defer to the administrator’s discretion”). “[Flexibility in the scope of judicial review need not require a proliferation of different standards of review; the arbitrary and capricious standard may be a range, not a point. There may be in effect a sliding scale of judicial review of trustees’ decisions.” Van Boxel v. Journal Co. Employees’ Pension Trust, 836 F.2d 1048, 1052 (7th Cir.1987); Chambers, 100 F.3d at 826 (quoting Van Boxel for same proposition). Therefore, we recognize that “the wholesale importation of the arbitrary and capricious standard into ERISA is unwarranted.” Firestone, 489 U.S. at 109, 109 S.Ct. 948. To say that there is a sliding scale of deference, however, merely begs the question: hoic much less deference ought a reviewing court afford? Our past opinions in this area do not clearly address this question. See, e.g., McGraw v. Prudential Ins. Co. of Am., 137 F.3d 1253, 1258 (10th Cir.1998) (recognizing the arbitrary and capricious standard as “inherently flexible” *1182 in that “the degree of deference ... will be decreased on a sliding scale in proportion to the extent of conflict” without providing guidance as to what a deference reduction entails). Our failure to articulate clearly the requirements of a less deferential arbitrary and capricious standard has left district courts in this circuit without direction and has encouraged litigation. 3 2. Defining the sliding scale: a two-tier approach In light of this lack of clarity, we capitalize on this opportunity to elaborate more fully what a less deferential standard of review entails. We adopt the following two-tiered standard for reducing deference in instances in which a fiduciary has a conflict of interest because it (1) adheres to ERISA common law, (2) promotes sound public policy, and (3) provides clearer guidance to lower courts, lawyers, and potential litigants. Under the first tier, in every case in which the plan administrator operates under a conflict of interest, the plan administrator bears the burden of proving the reasonableness of its decision pursuant to this court’s traditional arbitrary and capricious standard. Under the second tier, the court must determine whether the conflict of interest at issue is so severe as to warrant an additional reduction in deference. a. First tier Under the first tier, “if a benefit plan gives discretion to an administrator ... who is operating under a conflict of interest, that conflict must be weighed as a factor in determining whether there is an abuse of discretion.” Firestone, 489 U.S. at 115, 109 S.Ct. 948 (internal quotations omitted). When reviewing a plan administrator’s decision made under any conflict of interest, therefore, the reviewing court must afford the administrator less deference, see id., without placing such an onerous burden on the administrator as to make voluntarily provided benefits cost prohibitive. See Kathryn J. Kennedy, Judicial Standard of Review in ERISA Benefit Claim Cases, 50 Am. U.L.Rev. 1083, *1183 1087-88 (2001) (hereinafter, Kennedy, Judicial Standard). Professor Kennedy suggests a procedure for decreasing deference for first-tier cases that we find persuasive. [T]o further protect participants and beneficiaries in such conflict of interest contexts, [courts should] shift[] the burden to the fiduciary to justify the reasonableness of its decision. This puts the plan administrator on notice that its decisions will be judged for their reasonableness and provides the courts with a record that must show that the conflict of interest did not taint such decision. Such a result is still consistent with the Firestone admonition to consider as a factor any conflict of interest, but provides more direction for the courts in the application of the reasonableness standard. Id. at 1174. This burden shifting approach for first-tier cases has numerous advantages. First, it comports with our post-Firestone holdings that, even in cases of conflict of interest, the arbitrary and capricious standard provides the appropriate level of review. See Chambers, 100 F.3d at 827. Second, it provides clear direction to district courts, lawyers, and potential litigants. Third, it provides the less deferential review that we must accord a conflicted plan administrator without unduly raising insurance costs. See Sandoval v. Aetna Life & Cas. Ins. Co., 967 F.2d 377, 380 (10th Cir.1992) ("A primary goal of ERISA was to provide a method for workers and beneficiaries to resolve disputes over benefits inexpensively and expeditiously."). Thus, in all cases in which a plan administrator acted while subject to a conflict of interest, the plan administrator should bear the burden to prove the reasonableness of its actions, pursuant to our arbitrary and capricious standard of review. b. Second tier While not every type of conflict of interest will require further deference reduction, our previous jurisprudence requires that we "decreas[e] the level of deference in proportion to the severity of the conflict." Jones v. Kodak Med. Assistance Plan, 169 F.3d 1287, 1291 (10th Cir.1999). Thus, in certain types of cases merely shiftihg the burden ("first-tier" cases) will not adequately increase the level of judicial scrutiny in proportion to the severity of the conflict of interest. Therefore, in addition to shifting the burden of proof in all conflict cases, the reviewing court must grant even less deference to the plan administrator in those types of cases where a sufficiently serious conflict of interest exists ("second-tier" cases). Cf. Chambers, 100 F.3d at 827 ("{T]he arbitrary and capricious standard is sufficiently flexible to allow a reviewing court to adjust for the circumstances alleged, such as trustee bias in favor of a third-party or self-dealing by the trustee.") (quotation marks omitted). In particular, when a plan administrator, who also serves as the third party insurer, has denied coverage, an additional reduction in deference is appropriate: the denial of coverage may be upheld only if the plan administrator/third party insurer establishes by a preponderance of the evidence that the denial is warranted. See Kennedy, Judicial Standard, 50 Am. U.L.Rev. at 1174 (commenting on the shifting of the burden of proof to fiduciaries with conflicts of interest). Although we need not determine every instance in which a conflict of sufficient severity requires a further deference reduction, we do hold that such a sufficiently severe conflict exists where, as in the present case, the third-party insurer acts as the plan administrator and denies coverage for the insured. *1184 We note that at least two other courts of appeals, in considering this type of conflict, suggest that UNUM does not face an especially serious conflict of interest and requires only a small reduction in deference. For instance, the Fifth Circuit states in dicta: When a minimal basis for a conflict is established, we review the decision with only a modicum less deference than we otherwise would. In the instant case, the district court held that UNUM had an “inherent conflict of interest” because it was both the insurer and the plan administrator, which determined whether to pay claims under the Policy. Accordingly, the district court found that UNUM’s decision to deny Lain’s claim was subject to a “modicum less deference.” Lain v. UNUM Life Ins. Co. of Am., 279 F.3d 337, 343 (5th Cir.2002) (internal citations and some quotation marks omitted). The Fourth Circuit suggests a similar approach: In cases like this one, when the plan administrator is also the plan’s insurer, a conflict of interest exists. To safeguard against the potential for bias on the part of the insurer, a reviewing court must shift the standard of review to the extent necessary to counteract evidence of undue influence. We find virtually no evidence of self-dealing in the record, and therefore agree with the district court that the appropriate standard of review remains abuse of discretion, albeit with a concomitant skepticism of unsupported statements of UNUM representatives. Jones v. UNUM Life Ins. Co. of Am., 57 Fed.Appx. 159, 160 (4th Cir.2003) (unpublished) (citations omitted). We find neither Lain nor Jones persuasive. Here, the plan administrator directly profits from the denial of claims, due to its dual position as third-party insurer and plan administrator. Indeed, UNUM faces a more severe conflict of interest than that facing other conflicted plan administrators because UNUM is not subject to countervailing pressures that work in favor of insureds. For instance, a reduced conflict of interest exists in cases in which the employer, or the employee’s union, acts as the plan administrator, because they must balance their incentive to deny claims — and thus decrease premiums — with their desire to prevent declining employee morale, to retain skilled employees, and to mollify demands for wage increases. See Smathers v. Multi-Tool, Inc./Multi-Plastics, Inc. Employee Health & Welfare Plan, 298 F.3d 191, 197 (3d Cir.2002) (“[T]he risk of a conflict of interest is decreased where the administrator and funder of the plan is the employer, rather than an insurance company, because the employer has incentives to avoid the loss of morale and higher wage demands that could result from denials of benefits.”) (internal quotation marks omitted). The serious nature of the conflict of interest faced by third-party insurer/plan administrators such as UNUM, “puts the plan administrator on notice that its decisions will be judged for their reasonableness and provides the courts with a record that must show that the conflict of interest did not taint such decision.” Kennedy, Judicial Standard, at 1174. We thus modify our traditional arbitrary and capricious analysis in these cases to require support for the administrator’s decision to deny coverage by a preponderance of the evidence, rather than the traditional requirement of substantial evidence. Sandoval, 967 F.2d at 380 n. 4. Preponderance of the evidence presents a higher standard of proof than substantial evidence. Id. at 382 (holding that substantial evidence “is such evidence that a reasonable mind *1185 might accept as adequate to support the conclusion reached by the [decision maker] .... [and] requires more than a scintilla but less than a preponderance”) (internal quotation marks omitted). Application of this heightened standard of review comports with established law. 4 Under ERISA, an insurer bears the burden to prove facts supporting an exclusion of coverage. McGee v. Equicor-Equitable HCA Corp., 953 F.2d 1192, 1205 (10th Cir.1992). Federal courts treat insurer claims of policy exclusions as affirmative defenses. See 5 Charles Alan Wright & Arthur R. Miller, Federal Practice and Procedure: Civil 2d § 1271 (1990) (“[F]ederal courts have treated as [an] affirmative defense[ ] for purposes of Rule 8(c) ... [a] claim by an insurer that the loss suffered by the insured was excepted by the policy’s terms.”). Thus, as an affirmative defense, an exclusion of coverage requires that the insurer provide proof by a preponderance of the evidence. See, e.g., Brownlow v. Aman, 740 F.2d 1476, 1486-88 (10th Cir.1984) (holding that a defendant must prove its affirmative defenses by a preponderance of the evidence); Cleary v. Knapp Shoes, Inc., 924 F.Supp. 309, 315 (D.Mass.1996) (“[A] plan administrator attempting to establish exclusion from coverage has the burden to establish by a preponderance of evidence that a covered employee’s illness or medical condition is excludable.”) (internal quotation marks omitted). 3. The district court’s application of the standard of review Here, the district court noted that UNUM had admitted its inherent conflict of interest. Aplt’s App. at 12-13. The court then stated that it was “[fjully recognizing” that conflict. Id. at 13. UNUM argues that the district court’s use of the word “fully” in the order indicates that the district court engaged in the appropriate re-calibration along the sliding scale, while Ms. Fought argues that the district court did not reduce its deference in recognition of the conflict. We certainly do not question the district court’s awareness of a conflict. However, it is not completely clear how the district court’s full “recognition” of the conflict of interest affected its review of the plan administrator’s decision to deny benefits to Ms. Fought. Immediately after recognizing the conflict, the district court provided the following description of its analysis: “[T]he guiding inquiry must be whether the plan administrator’s decision was objectively reasonable given the administrative record — not whether a different reasonable decision could have been made.” Aplt’s App. at 13. This statement is difficult to distinguish from pure arbitrary and capricious deference: When reviewing under the arbitrary and capricious standard, [t]he Administra *1186 tor[’s] decision need not be the only logical one nor even the best one. It need only be sufficiently supported by facts within [his] knowledge to counter a claim that it was arbitrary or capricious. The decision will be upheld unless it is not grounded on any reasonable basis. Kimber v. Thiokol Corp., 196 F.3d 1092, 1098 (10th Cir.1999) (internal citations and quotation marks omitted). Notably, in Kimber, upon which the district court relied, see Aplt’s App. at 13, the court had considered and rejected the possibility that the administrator’s conflict of interest required it to slide away from the pure arbitrary and capricious standard. See Kimber, 196 F.3d at 1098 (“[T]here is insufficient evidence of a conflict of interest and review with deference is appropriate.”). The Kimber court’s “reasonable basis” language, echoed by the district court here, is thus not the proper inquiry in this ease. In Kimber, there was insufficient evidence of a conflict of interest. Here, the conflict was clear and uncontested. Under the standard we have set forth in this opinion, UNUM was required to justify its decision to exclude coverage by a preponderance of the evidence. Under that standard, we now proceed to consider the district court’s grant of summary judgment to UNUM. In so doing, we examine the language of the plan and the evidence on which UNUM relies in denying Ms. Fought’s claims for benefits. B. The Plan’s Language The policy at issue here, as noted above, includes the following language relevant to this appeal: “Your plan does not cover any disabilities caused by, contributed to by, or resulting from your ... pre-existing condition.” Aple’s Supp.App. at 341-42. You have a pre-existing condition when you apply for coverage when you first become eligible if: - you received medical treatment, consultation, care or services including diagnostic measures or took prescribed drugs or medicines in the 3 months just prior to your effective date of coverage; or you had symptoms for which an ordinarily prudent person would have consulted a health care provider in the 3 months just prior to your effective date of coverage; and - the disability begins in the first 12 months after your effective date of coverage. Id. at 342. Applying our more searching and less deferential standard of review in light of UNUM’s admitted conflict of interest, we must determine whether UNUM established by a preponderance of the evidence that Ms. Fought’s claim was not covered by the plan. “We are mindful that the objective in construing a health care agreement, as with general contract terms, is to ascertain and carry out the true intention of the parties. However, we do so giving the language its common and ordinary meaning as a reasonable person in the position of the [plan] participant, not the actual participant, would have understood the words to mean.” Pitman, 217 F.3d at 1298 (quoting Blair v. Metropolitan Life Ins. Co., 974 F.2d 1219, 1221 (10th Cir.1992) (internal quotation marks omitted)). 1. The role of causation in interpreting the pre-existing condition clause “Cause” means “[t]o be the cause of,” which is “[something that produces an effect, result, or consequence.” Webster’s II New Riverside University Dictionary 239 (1988). “Contributed” is defined broadly as “[t]o act as a determin *1187 ing factor.” Id. at 306. “Results” means “to happen or exist as a result of a cause.” Id. at 1002. Ms. Fought argues that UNUM imper-missibly extended the language of the policy such that it excludes coverage for disabilities that result from surgery, not those that result from pre-existing conditions. The major difficulty presented by this case is that UNUM’s policy excludes coverage for disabilities caused by preexisting conditions, whereas it seeks here to apply its policy as if it excludes coverage for disabilities caused by complications from surgery for pre-existing conditions. Surgery is not, of course, a preexisting condition, but at most a necessary consequence of a pre-existing condition. In essence, therefore, this case becomes a matter of where we draw the line on chains of causation. UNUM responds that the broad language of the pre-existing condition dictates a similarly broad interpretation of the exclusion: the exclusion does not require that the disabling condition be the sole or direct result of the pre-existing condition. Here, UNUM applies the limitation because it believes the disabling condition was “caused by, contributed to, or resulted from” Ms. Fought’s pre-existing condition. Based on the common ordinary meaning of the terms “cause,” “contribute,” and “result,” UNUM contends, the “exclusion merely requires that [Ms.] Fought’s preexisting heart condition be ‘something’ that brought about the disabling condition or that played a significant part in bringing about the disabling condition, or that the disabling condition arose as a consequence of the pre-existing condition.” Aple’s Br. at 25. In practice, however, UNUM’s arguments rely upon classic but/for causation: But for the coronary artery disease, none of the rest of the chain of events would have happened. Or, as Ms. Fought herself put it: “It is kind of like saying Tf I hadn’t went outside in the rain, I wouldn’t have got struck by lightening, [sic]’ ” Aplt’s App. at 157 (Letter to New Mexico Public Relations Commission, dated Jan. 31, 2000). As Ms. Fought persuasively argues, the chain of non-proximate causation that UNUM asserts in her case is attenuated to the point of absurdity: [UNUM’s] argument necessarily goes something like this: but for the preexisting coronary artery disease, Ms. Fought probably would not have the surgery; but for the surgery, Ms. Fought would not have had a surgical wound; but for the surgical wound, Ms. Fought’s previously undetectable osteo-porotic sternum would not have prevented her doctors from closing her wound in a more conventional manner, which might have given the wound greater stability and resistence to the lateral tension in the wound exerted by Ms. Fought’s large breasts; but for the combination of the surgical wound, Ms. Fought’s osteoporotic sternum and her large breasts, the wound probably would not have dehisced, thereby providing an entry point for the staph infection several weeks after the surgery; and, but for the fact that the staph infection was resistant to antibiotics, entered Ms. Fought’s bloodstream, and eventually spread to other parts of her body, Ms. Fought would not be disabled. Aplt’s Br. at 21-22. While the steps of causation are undoubtedly drawn out for effect, the larger point is a valid one. For Ms. Fought, there were at least five intervening stages between the pre-existing coronary artery disease and the disability: The failure of non-surgical alternatives, initially successful elective surgery, later complications from that surgery, initially successful *1188 treatment of those complications, and finally a drug resistant infection due to those complications, which in itself may have been caused by the intervening presence of Staphylococcus aureus due to faulty sterilization, sanitation, etc. UNUM seems to suggest that it need not cover anything for which it can construct a but/for story. If we were to accept this contention, we would effectively render meaningless the notion of the pre-existing condition clause by distending the breadth of the exclusion. 2. Department of Labor’s regulations and example The Department of Labor’s regulations also undermine UNUM’s interpretation of the plan. Those regulations provide the following example regarding the scope of a pre-existing condition exclusion: 5 Example 4. (i) Individual D, who is subject to a preexisting exclusion imposed by Employer U’s plan, has diabetes, as well as a foot condition caused by poor circulation and retinal degeneration (both of which are conditions that may be directly attributed to diabetes). After enrolling in the plan, D stumbles and breaks a leg. (ii) In this Example 4, the leg fracture is not a condition related to D’s diabetes, even though poor circulation in D’s extremities and poor vision may have contributed towards the acci- . dent. However, any additional medical services that may be needed because of D’s preexisting diabetic condition that would not be needed by another patient with a broken leg who does not have diabetes may be subject to the preexisting condition exclusion imposed under Employer U’s plan. 29 C.F.R. § 2590.701 — 3(a)(l)(i)(C), Example 4. In addition, the Practicing Law Institute provided these guidelines in interpreting § 2590.701-3: Thus, before imposing a preexisting condition limitation, plan sponsors must carefully evaluate whether a particular condition is “directly attributable” to the preexisting condition. Medical conditions which merely “contribute towards” accidents or illnesses, but are not “directly attributable” to the preexisting condition may not be excluded. This causal connection requirement will undoubtedly open the door for arguments that preexisting conditions were not the “proximate cause” of a particular injury or sickness — e.g., treatment of pneumonia for an individual who was previously diagnosed with AIDS. John R. Hickman, Insurance Law: What Every Lawyer and Businessperson Needs to Know, Health Insurance Basics: ERISA, FMLA, ADA, ADEA, COBRA, HIPAA, and PARCA, 584 PLI/Lit 413, 487 (May 1998). When applying the above example from § 2590.701-3 to Ms. Fought’s case, we determine that UNUM’s expansive reading of the exclusion may be overly broad: The exclusion cannot merely require that the pre-existing condition be one in a series of factors that contributes to the disabling condition; the disabling condition must be substantially or directly attributable to the pre-existing condition. See also Webster’s II New Riverside Dictionary 306 (defining contribute as to mean “to act as a determining factor”). *1189 Ms. Fought’s staph infection is not a condition related to her coronary artery disease, even though her unstable angina, which was related to her coronary artery disease, undoubtedly contributed to the need for surgery. To read the exclusion as broadly as UNUM, counters the essential tenets of contract law: Exclusions must be interpreted narrowly. See 29 C.F.R. § 2590.701-3(a)(l)(i)(C), Example 4 (applying narrow definition of “contributed towards”); Frerking, 760 F.Supp. at 881 (noting that in the context of plans governed by ERISA, “[i]t is also well-established that the burden is upon the insurer to demonstrate that the insured’s claim falls within the terms of an exclusionary clause, and that such clauses are interpreted narrowly”). 3. Illustrative cases The few cases that have focused on the application of proximate cause to exclusions for pre-existing conditions are not inapposite. In Cash v. Wal-Mart Group Health Plan, 107 F.3d 637 (8th Cir.1997), the plaintiff was diagnosed with diverticu-lar disease. He later developed diverticulitis. The Eighth Circuit held the denial of benefits was reasonable, because the diverticulitis was a complication and secondary condition of the presence of diverticula in the wall of the colon. Id. at 643. In other words, the diverticular disease was a “necessary precursor” to the later illness of diverticulitis. Id. Here, in contrast, Mr. Fought’s coronary condition was not a prerequisite to the onset of the staph infection: There is no necessary precursor link. Similarly, in Holsey v. UNUM Life Ins. Co. of Am., 944 F.Supp. 573, 579 (E.D.Mich.1996), relied upon by the district court, “[Plaintiffs] blindness was caused by, contributed to by or resulted from diabetes where blindness was related to diabetes and glaucoma was a well-known complication of diabetes; preexisting condition exclusion enforced.” Aplt’s App. at 14. Holsey, like Cash, clearly describes a situation where an insurer denied coverage for the results of diabetes, not for the complications from treatment or su/rgery for diabetes. Blindness certainly is a well-known complication of diabetes. Staph infections are not, so far as we are aware, a well-known complication of coronary artery disease. The district court also cited to Currie v. Metropolitan Life Ins. Co., No. CIV-A-1665, 1998 WL 214761, at *3-*4 (E.D.La. April 29, 1998), and noted the Currie court “rejected the] argument that plaintiff was suffering from a different sickness or injury when she received treatment within the pre-existing period; all treatment stemmed from prior car wreck; preexisting condition exclusion enforced.” Aplt’s App. at 14. The plaintiff in Currie was disabled because of back pain. She had been treated for back pain after a car accident that had occurred before she was covered by the insurance policy, so the insurer denied coverage for a pre-existing condition. She tried to claim that she was at that point suffering from a different kind of back pain, whereas the back pain for which she had received treatment was from a different cause; but the court held that all of the back pain had arisen from the car accident. This scenario, however, has nothing in common with Ms. Fought’s case. The question is not whether her disability arose from a different surgery or a different form of heart disease but simply whether the causal connection between the disease and the disability is insufficiently proximate. Next, in Reinert v. Giorgio Foods, Inc., 15 F.Supp.2d 589 (E.D.Pa.1998), the plaintiff, Ms. Giorgio, suffered from three pre-existing conditions before she became eligible for plan benefits: diabetes, Char-cot joint disease, and diabetic neuropa-thy. She later suffered from a series of ulcerations on her left foot. Two of the ulcerations were manifestations of her *1190 pre-existing conditions. A third ulceration developed from an insulin needle that became embedded in the plaintiffs foot. The needle ulceration was a “separate and distinct injury which was aggravated by those underlying conditions.” Id. The court found improper the denial of benefits for the treatment of this injury. The court recognized, however, that the distinction between the ulcers was subtle. Ulcerations similar to those suffered by Ms. Giorgio were frequently caused by continuing deterioration of the bones and tissue in her foot, which was in turn caused by the diabetes, Charcot joint disease, and diabetic neuropathy. Thus, applying a strict arbitrary and capricious standard, with no conflict of interest present, the court determined that the improper denial of benefits as to the third ulcer was not arbitrary and capricious. Here, UNUM cannot point to such nuances. The staph infection was a separate and distinct injury, not a manifestation of the underlying coronary disease. Finally, and most importantly, in Vander Pas v. UNUM Life Ins. Co. of Am., 7 F.Supp.2d 1011 (E.D.Wis.1998), the plaintiff had suffered from a pre-existing heart condition for which he took the drug Coumadin. Coumadin puts a patient at risk for a subdural hematoma. The patient did suffer a subdural hematoma, and UNUM attempted to deny coverage on the theory that the pre-existing heart condition had caused him to take Coumadin, which had then caused the hematoma. The district court in Vander Pas described the chain of causality as “attenuated: [T]he plaintiffs atrial fibrillation caused him to take Coumadin, which brought about his subdural hematoma, which produced his disability.” id. at 1018. The court faulted UNUM for not providing a “proximate cause analysis,” among other failings. Id. The district court therefore denied UNUM’s motion for summary judgment. UNUM attempts to distinguish Vander Pas by asserting that the court did not actually rule against UNUM on the basis of lack of proximate cause, but because UNUM “had not set forth that, or any, explanation of a chain of causation.” Aple’s Br. at 15 (bold-faced type in original). This misreads Vander Pas. The court clearly based its decision on UNUM’s failure to explain how the treatment itself, Coumadin, was the “pre-existing condition.” Vander Pas, 7 F.Supp.2d at 1018. In other words, the district court properly required UNUM to show that the proximate cause of the disability (taking Coumadin) pre-existed. That the patient would not have been taking Coumadin but for the pre-existing heart condition did not make the pre-existing condition the “cause” of the disability. See id. (“[T]he proposition that Coumadin played a part in causing plaintiffs subdural hematoma ... is not equivalent to a studied conclusion that plaintiffs use of Coumadin satisfies the definition for ‘pre-existing condition,’ or that his disability was ‘caused by, contributed to by, or resulted] from’ the use of Coumadin.”). Similarly here, UNUM must demonstrate that the proximate cause of the disability, here, the staph infection, was a pre-existing condition. 4. Availability of clearer language Ms. Fought next argues that there is certainly a clearer way to write a contract that would exclude coverage for complications from surgery. Indeed, she notes a district court case from this circuit that considered a contract with precisely that language: “Pursuant to the Plan, long-term disability benefits ... are not payable for any[ ] ... [disability caused or contributed to by a Preexisting Condition or medical or surgical treatment of a *1191 Preexisting Condition.” Kaus v. Standard Ins. Co., 985 F.Supp. 1277, 1279 (D.Kan.1997) (emphasis added), aff'd, No. 97-3378, 1998 WL 778055 (10th Cir. Nov.5, 1998); see also Reinert, 15 F.Supp.2d at 595 (suggesting the redrafting of the exclusion to encompass “all conditions relating to pre-existing illnesses” so as to avoid similar claims). In addition, the exclusion might be drafted to cover bacterial infections. See, e.g., Lewin v. Metropolitan Life Ins. Co., 394 F.2d 608, 609 (3d Cir.1968) (applying accident insurance policy). The district court acknowledged that the language “may be clearer in Kaus ” but it reasoned that the contract language here is “broad enough to encompass the treatment received in this case.” Aplt’s App. at 15. UNUM is a sophisticated party, and the plan’s language is, on the whole, careful and thorough. The existence of policies, as in Kaus, that specifically deny coverage on the basis of complications from surgery lends support to the argument that it is unreasonable as a matter of law to conclude that the general language in Ms. Fought’s contract encompasses the same result. UNUM had every opportunity to add the words “or medical or surgical treatment of a Preexisting Condition,” Kaus, 985 F.Supp. at 1279, but it did not do so. It is unreasonable to allow it to do so post facto,. to the detriment of Ms. Fought and other insureds. Given the Department’s regulations invoking proximate cause, the illustration above, and the availability of clearer language, we conclude that the language of the exclusion provision in the contract should not be extended to the degree that UNUM attempts to do here. See 29 C.F.R. § 2590.701 — 3(a)(l)(i)(C), Example 4; Frerking, 760 F.Supp. at 881 (citing caselaw requiring a narrow interpretation of exclusion clauses). C. Sufficiency of the evidence Our inquiry does not end here, however. We still must determine whether the improper denial of benefits was supported by a preponderance of the evidence. In denying disability benefits to Ms. Fought, UNUM relied in part on the discharge report presented by Dr. Waljii. Dr. Waljii’s report indicated that, during the heart bypass surgery, he discovered that Ms. Fought’s sternum was “narrow” and “very osteoporotic.” Aplt’s App. at 79. Dr. Waljii attempted to “reapproximate the sternum,” using wires that tore through it. Id. He removed the wires and reapproximated the sternum using a surgical procedure called “Robichek reinforcement.” Id. The remainder of the wound was closed in a “standard three layer closure.” Id. He reported: Because of the large and pendulous breasts as well as significant adiposity, the lower portion of the skin incision was also at risk of dehiscence as indeed was the full sternum. Clearly her postoperative course was anticipated to be quite challenging given the concerns about the wound as indeed her other medical problems. Id. Ms. Fought returned to the hospital a few weeks later for “wound care of this sternal dehiscence and quite possibly ster-nal infection.” Aplt’s App. at 53. She was sent to a nursing facility a few days later. Approximately three weeks later she was readmitted because of increased wound pain. After two exposed sternal wires were detected, the methicillin-resistant Staphylococcus aureus infection was diagnosed in her bloodstream. Ms. Fought’s initial application for long term disability, filed in June 1999, while she was hospitalized, indicates that she was unable to work since the time of her *1192 open heart surgery. Her application’s physician statement indicated that she had coronary artery disease, that she underwent bypass surgery, and that she continued to be hospitalized due to the sternal wound infection. As described above and as the medical records indicate, Ms. Fought was hospitalized for two months and underwent several operations and procedures as a result of the inability to clear up the staph infection. UNUM’s in-house pre-existing medical review concluded that the conditions were “most likely caused by, contributed to, or resulted from [Ms. Fought’s previously diagnosed severe hypertension and mod/severe left ventricle hypertrophy.]” Aplt’s App. at 72. The in-house review took one day to complete and yielded a one-paragraph opinion that indicated that “[m]edical records from this period could further strengthen this opinion.” Id. UNUM acknowledges that the staph infection was not present during the pre-existing time period, but that it resulted from the surgery “that was performed for a cardiac condition that was present ... during that time frame.” Aplt’s App. at 147. In support of her argument, Ms. Fought presents notes from three doctors indicating that the staph infection was a separate condition, unrelated to the coronary artery disease, without a preceding history. See Aplt’s App. at 156. (Note from Dr. Robert T. Ferraro, dated Oct. 19, 1999) (“[T]he staph infection which is the basis for multiple wounds on chest is not related to coronary artery disease. This is a separate, unrelated diagnosis without preceding history.”); id. at 154 (Note from Dr. Robert Dubroff, dated Oct. 22, 1999) (“[Ms.] Fought is totally disabled due to her heart condition. The staph infection was not a pre-existing condition.”); id. at 155 (Note from Dr. Neil T. Chen, undated) (“[Ms.] Fought’s chest and abdominal wounds/infection is [sic] not a preexisting condition.”). (emphasis in original). Two of the three physicians were treating physicians. UNUM counters that the notes do not state that the staph infection caused her disabling condition, in fact, one note indicates that the disability was “due to her heart condition.” Id. at 154. UNUM thus contends its reliance on the medical records, in conjunction with the medical notes, supports the plan administrator’s conclusion that it was the coronary artery disease that set in motion the staph infection. We acknowledge that, in some instances, the presence of conflicting evidence in the record may establish that a plan administrator’s decision was reasonable. See Sandoval, 967 F.2d at 382. However, here, applying the new standard, UNUM must establish by a preponderance of the evidence that its denial of benefits was warranted by the plan language and the evidence. Moreover, the only arguably conflicting evidence here is the note from Dr. Dubroff that states both that Ms. Fought may be disabled “due to her heart condition” and that her “staph infection was not a pre-existing condition.” Aplt’s App. at 154. The bulk of Ms. Fought’s medical records indicate that she was recovering well from the surgery; the onset of the staph infection, which originated from an unknown origin, was the debilitating condition. UNUM is unable to offer “more than a scintilla” of evidence, Sandoval, 967 F.2d at 382, that the staph infection was a manifestation of the pre-existing coronary artery condition, that it was caused by the pre-existing condition, or that it was substantially contributed to by or resulted from the pre-existing condition. In addition, despite its apparent and admitted conflict of interest, UNUM undertook no independent evaluation or investigation. We note that, while not required, independent medical examinations are of *1193 ten helpful. Where a conflict of interest exists, the Seventh Circuit encourages, if not requires, such an inquiry: “When it is possible to question the fiduciaries’ loyalty, they are obliged at a minimum to engage in an intensive and scrupulous independent investigation of their options to insure that they act in the best interests of the plan beneficiaries.” Hightshue v. AIG Life Ins. Co., 135 F.3d 1144, 1148 (7th Cir.1998) (internal quotation marks omitted). “Seeking independent expert advice is evidence of a thorough investigation.” Id.; see also Woo v. Deluxe Corp., 144 F.3d 1157, 1161 (8th Cir.1998) (holding that administrator erred in failing to obtain an independent review of the claim by an expert when treating physician determined applicant was disabled and when there was evidence of an uncommon disease). Where, as here, a conflict of interest may impede the plan administrator’s impartiality, the administrator best promotes the purposes of ERISA by obtaining an independent evaluation. See 29 U.S.C. § 1001(b). In sum, we cannot say that, given UNUM’s conflict of interest, and our modified sliding scale review, that UNUM has justified its denial of benefits by a preponderance of the evidence. UNUM’s repeated denials of Ms. Fought’s claims merely echo the terms of the pre-existing condition plan language, and offer at most thin support from the record before us. III. CONCLUSION We thus hold that the plan’s language here does not reasonably apply to the attenuated chain of events between Ms. Fought’s pre-existing coronary artery disease and her disabling staph infection and that UNUM’s denial of benefits was not supported by a preponderance of the evidence. Accordingly, we REVERSE the district court’s grant of summary judgment to UNUM and REMAND for further proceedings consistent with this opinion. 6 Footnotes [1] The letter from UNUM does not mention the other doctors’ letters, saying only , that "Dr. Ferraro’s note was reviewed by our medical department.” Aplt’s App. at 147. Apparently, the other two doctors’ letters were faxed to UNUM on October 25, 1999, the very day that UNUM sent Ms. Fought the letter confirming its denial of disability benefits. See id. at 146. "Plan administrators, of course, may not arbitrarily refuse to credit a claimant’s reliable evidence, including the opinions of a treating physician.” Black & Decker Disability Plan v. Nord, 538 U.S. 822, 123 S.Ct. 1965, 1972, 155 L.Ed.2d 1034 (2003). We note that UNUM's subsequent review, see Aplt's App. at 150, appeared to include consideration of Ms. Fought’s additional doctors’ notes. We also note that the parties do not argue that the plan administrator misallocated the weight it gave to the treating physician’s opinion, so the admonishments of Black & Decker do not apply. See Black & Decker, 123 S.Ct. at 1972 ("[W]e hold [that] courts have no warrant to require administrators automatically to accord special weight to the opinions of a claimant’s physician; nor may courts impose on plan administrators a discrete burden of explanation when they credit reliable evidence that conflicts with a treating physician’s evaluation.”). [2] We continue to treat the terms "arbitrary and capricious" and "abuse of discretion” as interchangeable in this context. See Chambers v. Family Health Plan Corp., 100 F.3d 818, 825 n. 1 (10th Cir.1996) ("Some circuit courts have recently distinguished between these two standards and have concluded that the abuse of discretion standard is more appropriate. Most courts, however, have held that this is a distinction without a difference. We agree and adhere to the arbitrary and capricious standard of review.”) (internal citations and quotation marks omitted). [3] Our review of the law reveals a similar lack of direction from other circuits. See, e.g., Pinto v. Reliance Standard Life Ins. Co., 214 F.3d 377, 390-93 (3d Cir.2000) (holding, without providing further guidance, that the deference accorded the fiduciary will be lessened by the degree necessary to neutralize influence resulting from conflict); Vega v. Nat’l Life Ins. Servs., Inc., 188 F.3d 287, 296 (5th Cir.1999) (stating only that the "deference [provided to the administrator] will be lessened to the degree necessary to neutralize any untoward influence resulting from the conflict”); Borda v. Hardy, Lewis, Pollard & Page, P.C., 138 F.3d 1062, 1069 (6th Cir.1998) (concluding that the fiduciary will be entitled to some deference, but that "application of the standard should be shaped by the circumstances of the inherent conflict of interest”); Mers v. Marriott Int’l Group Accidental Death & Dismemberment Plan, 144 F.3d 1014, 1020 n. 1 (7th Cir.1998) (recognizing that "[t]he arbitrary and capricious standard does not pose an all-or-nothing choice between full deference or none”); Woo v. Deluxe Corp., 144 F.3d 1157, 1161-62 (8th Cir.1998) (describing the sliding scale as extremely flexible in that courts may adjust for all relevant circumstances). The Third Circuit states this well: We acknowledge that there is something intellectually unsatisfying, or at least discomforting, in describing our review as a "heightened arbitrary and capricious" standard. The locution is somewhat awkward. The routine legal meaning of an "arbitrary and capricious” decision is that used, quite understandably, by the district court: a decision "without reason, unsupported by substantial evidence or erroneous as a matter of law.” Once the conflict becomes a "factor” however, it is not clear how the process required by the typical arbitrary and capricious review changes. Pinto, 214 F.3d at 392. [4] There is correlating line of authority which places the burden on the insurer to establish that the denial of benefits fell within the narrowly construed exclusionary clause. See Caffey v. UNUM Life Ins. Co., 302 F.3d 576, 580 (6th Cir.2002) ("ERISA places the burden of proving an exclusion from coverage in an ERISA-regulated welfare plan on the plan administrator.”); Frerking v. Blue Cross-Blue Shield of Kan., 760 F.Supp. 877, 881 (D.Kan.1991) (noting that "[i]t is also well-established that the burden is upon the insurer to demonstrate that the insured's claim falls within the terms of the exclusionary clause, and that such clauses are interpreted narrowly”) (citing Tex. E. Transmission Corp. v. Marine Office-Appleton & Cox Corp., 579 F.2d 561, 564 (10th Cir.1978) and Milliken v. Fidelity & Cas. Co., 338 F.2d 35, 41 & n. 13 (10th Cir.1964)); Cleary v. Knapp Shoes, Inc., 924 F.Supp. 309, 315 (D.Mass.1996) (noting that "it is a general rule of insurance law that the insurer bears the burden of showing that a covered injury falls within an exclusion provision”) (citing McGee v. Equicor-Equitable HCA Corp., 953 F.2d 1192, 1205 (10th Cir.1992)). [5] Those regulations generally provide that "a group health plan, and a health insurance issuer offering group health insurance coverage, may impose, with respect to a participant or beneficiary, a preexisting condition exclusion only if the requirements of this paragraph (a) are satisfied.” 29 C.F.R. § 2590.701-3(a)(l)(i). Those regulations include the length of time during which the condition was treated and the kind of treatment that has been received for the condition. See id. [6] Because we reverse the district court's grant of summary judgment in favor of UNUM, we also vacate the district court's order that UNUM recover its costs of action from Ms. Fought.